no gene_id for tRNA in GFF3

Question

[prasundutta87]

Hi,

In one of the non-human organisms, GFF3 annotation file downloaded from NCBI genome FTP, there were some tRNA annotations (tRNA-Glu, tRNA-Ser, etc) in the mitochondrial genome that did not have a gene_id. They don't have any transcript features associated with them as well. Is it a usual annotation issue in the GFF3 format or by practice tRNA genes don't have either gene_id or transcripts associated with them?

The problem is that when the GFF3 is converted to GTF using gffread, no gene_id is present for those tRNAs and hence creating downstream issues. Should they be removed altogether if we are not exclusively interested in them?

Answer 1

[lieven.sterck]

GFF(3) is the worst 'standard' , so yes unfortunately all different flavors and values being present or not can occur in a gff3 file.

I must add that the tRNAs (well all non mRNAs actually) is a bit of a grey zone, some will add the gene_id or even transcript to them and other providers will not

Answer 2

[prasundutta87]

Hmm..ok..so its better to just drop them off in our case.

[lieven.sterck]

if you're not interested in them , sure you can omit them

[prasundutta87]

I was also thinking of an option to give dummy gene and transcript IDs to them.. but haven't done that yet..can that be a workaround as well?

[lieven.sterck]

that's an alternative indeed. if you don't really need them you can 'fix' your gff file until it's valid

or perhaps use the gffvalidator which can try to fix some errors

[prasundutta87]

Thanks a lot Lieven! This was very helpful.

[lieven.sterck]

you're welcome!

if this has resolved your issue would you mind flagging this one as 'solved'? (probably need to convert one of the posts as an answer though)

[prasundutta87]

I could only moderate my post to make it as an 'answer'..the main answer should be your first reply..

[lieven.sterck]

right, indeed, only I can move my own posts up.